A group-III nitride based compound semiconductor is a direct transition type semiconductor, which is used as a light-emitting device, for example, has a wide range light-emission spectrum covering ultraviolet to red, and it is applied to light-emitting devices such as a light-emitting diode (LED) and laser diode (LD). This type of light-emitting device has a higher light emission efficiency compared to a conventional illumination-related light source, and therefore a low level of energy consumption thereof can be achieved. Moreover, this type of light-emitting device also has an advantage of long life, and the market for blue-color LEDs in particular is rapidly expanding.
This type of group-III nitride semiconductor light-emitting device is manufactured generally by being film-formed on a substrate composed of, for example, sapphire (Al2O3) by means of a metal-organic chemical vapor deposition (MOCVD) method with raw materials including trimethyl gallium, triethyl aluminum, and ammonia. In the MOCVD method, vapors of the materials contained in a carrier gas are transported to the substrate surface, and the materials are decomposed on the heated substrate surface, to thereby grow crystals.
Conventionally, single crystal wafers of group-III nitride semiconductor have not been commercially available, and group-III nitride semiconductors are commonly obtained by growing crystals on a single crystal wafer of a different material. However, between such different types of substrates and group-III nitride semiconductor crystals epitaxially grown thereon, there is a considerable lattice misfit. For example, in a case where gallium nitride (GaN) is grown on a sapphire substrate, 16% lattice misfit is present between them. In general, if such a considerable lattice misfit is present as described above, it becomes difficult to epitaxially grow crystals directly on the substrate.
Consequently, there has been proposed and commonly practiced a method in which when epitaxially growing group-III nitride crystals on a sapphire single crystal substrate or on a SiC single crystal substrate by means of the MOCVD method, first, a layer called a low temperature buffer layer composed of aluminum nitride (AlN) or aluminum gallium nitride (AlGaN) is laminated on the substrate, and group-III nitride semiconductor crystals are epitaxially grown thereon at a high temperature (for example, refer to Patent Documents 1 and 2). According to the method disclosed in Patent Documents 1 and 2, by growing group-III nitride semiconductor crystals on a low temperature buffer layer formed on a substrate, it is possible to form single crystal GaN with superior crystallinity and obtain a light-emitting device having a high level of brightness.
On the other hand, although the type of technique disclosed in Patent Documents 1 and 2 which uses a low temperature buffer layer enables formation of signal crystal GaN, this type of GaN single crystals have higher defect density and the level of crystallinity thereof is lower compared to single crystals of other materials. However, in general, in a group-III nitride semiconductor light-emitting device composed of GaN, light emission output can be obtained at or above a certain level even in those cases where the level of crystallinity thereof is lower to a certain degree. Therefore, defect density has been conventionally considered as having almost no impact on light emission output as long as crystallinity is at or above a certain level. For example, it has been thought that in a light-emitting layer which is provided in a group-III nitride semiconductor light-emitting device and which has a multiple quantum well structure formed with GaN barrier layers and InGaN well layers, there is almost no correlation between photoluminescence (PL) and threading dislocation density, and light emission output is not dependent on threading dislocation density. For this reason, conventionally, as research of a technique for improving light emission efficiency of a group-III nitride semiconductor device, there has been mainly researched a technique for improving light extraction efficiency, rather than a technique for increasing crystallinity and improving internal quantum efficiency to thereby increase light emission intensity. Here, light extraction efficiency refers to a ratio of light which can be extracted to the outside of the light-emitting device to the light generated in the light-emitting layer.
In general, examples of an indicator of light emission efficiency of a light-emitting device include internal quantum efficiency described above as well as external quantum efficiency. Here, internal quantum efficiency is the proportion of electric current energy supplied to the device, which is converted into light in the light-emitting layer. Moreover, external quantum efficiency refers to a value obtained by multiplying the internal quantum efficiency by the light extraction efficiency, and may serve as an indicator of how effectively the generated light can be used. That is to say, a light-emitting device may be referred to as a light-emitting device of high light emission output if the external quantum efficiency level thereof is high. Furthermore, it is necessary to improve light extraction efficiency in order to improve external quantum efficiency.
Therefore, in recent years, as a method for improving light emission efficiency of a light-emitting device, there has been mainly researched a technique for increasing light extraction efficiency to thereby improve external quantum efficiency. Moreover, based on the conventional concept in which there is no clear correlation between crystallinity and internal quantum efficiency, it has been thought that improvement in light emission efficiency by improving internal quantum efficiency was nearing the limit.
Here, an indicator of the crystallinity of a group-III nitride semiconductor composed of GaN or the like may be obtained, for example, by conducting X-ray diffraction on p-GaN (p-type semiconductor layer) which serves to be the topmost layer and which forms an LED structure, to thereby find an FWHM (full width at half-maximum) of the X-ray rocking curve of the (0002) plane and (10-10) plane of the p-GaN.
For example, crystallinity of a buffer layer composed of AlN, GaN, or the like mentioned in Patent Documents 1 and 2 above, is of the order of approximately several thousands arcsec to several ten-thousands arcsec in terms of FWHM. Therefore, it has been conventionally thought that even in a case where the crystallinity of an upper layer improves as respective layers constituting the LED structure are further grown on this buffer layer, the limit of the FWHM of p-GaN of the topmost layer is approximately 100 arcsec for the (0002) plane and approximately 300 arcsec for the (10-10) plane.
Moreover, it has been thought that, for example, when GaN crystals with a (0002) plane of 100 arcsec and a (10-10) plane of 300 arcsec in terms of FWHM are compared with GaN crystals with a (0002) plane of 150 arcsec and a (10-10) plane of 380 arcsec, if both of them are provided with an LED structure, growth conditions for growing the LED structure have a greater impact on light emission efficiency than that of crystallinity, and there is almost no correlation with crystallinity.
After the invention of a low temperature buffer, crystallinity of GaN single crystals has improved considerably, and realization of p-type semiconductor GaN as well as realization of a double-heterostructure LED, which uses InGaN for its active layer, have improved the brightness of an LED of UV, blue, and green, consequently expanding a new market. However, while crystallinity has improved, the GaN crystallinity still has a defect density of approximately 10+9/cm2, and this defect density is extremely high, compared to a GaAs based epitaxial single crystal thin film, which achieves approximately 1×10+4/cm2. Since light-emitting devices with 10% or more of external quantum efficiency have started to be made despite this, much research is now being conducted with a particular interest in finding out the reason why a GaN based LED can achieve a high level of light emission efficiency despite its high defect density.
One of the reasons why this type of research is being conducted is that improvement in crystallinity has reached its limit. In application for a laser diode, crystallinity needs to be further increased since there is a correlation between threshold current and crystallinity, and an epitaxial lateral overgrown method has been employed as a method for this. However, while this method partially improves crystallinity, it does not improve crystallinity of the entire wafer surface. Therefore, application of this method has not been preferred for an LED, in which cost is an important factor and the number of devices obtained from a single wafer is important, while the method may be applied to a laser diode. Consequently, the main interest in the area of increasing the brightness level of an LED has been brought to light extraction efficiency improvement, and there is a decreasing trend in the amount of research in the area of internal quantum efficiency improvement achieved by improving crystallinity.
Here, in a GaP based green LED, increased dislocation has been confirmed when conducting a TEM observation on a degraded chip during light emission (for example, refer to Non-Patent Document 1). Moreover, in a ZnSe based blue-green LED, it has been confirmed by TEM observation, and reported, that heretofore existing lamination defects grow and small dislocations are generated at a high level of density, creating non light-emitting centers (for example, refer to Non-Patent Document 2). In these cases of Non-Patent Documents 1 and 2, the above problem occurs even if the dislocation density is at a level of 10+4/cm2. Here, unwanted energy hierarchy is established if a dislocation is present, and if excited electrons fall thereinto, light is no longer emitted at its original wavelength. Therefore, this type of state is referred to as a non light-emitting center. In a GaN based LED, light emission efficiency can be obtained at or above 10% even if the dislocation density is 10+10/cm2, and the lifetime thereof can achieve a longer time compared to that of a GaAs based LED. Therefore, dislocations have not been considered as being the non light-emitting centers. In contrast, Sugahara et al. (for example, refer to Non-Patent Document 3) and Rosner et al. (for example, refer to Non-Patent Document 4) conducted a TEM observation on the same location where CL light emission took place to thereby prove that dislocations are the non light-emitting centers also in a GaN based LED.
Here, Mukai and Nakamura et al. reported in 1998 that in manufacturing a prototype of an LED which emits light at a peak wavelength in the vicinity of 370 nm, when active layers composed on InGaN slightly containing In and GaN used as an active layer are compared in a structure of p-GaN 1200 Å/p-AlGaN 600 Å/undoped InGaN 400 Å/n-AlGaN 300 Å/n-GaN 4 μm/GaN Buffer 300 Å, the level of light emission output is 10 times higher when using InGaN (for example, refer to Non-Patent Document 5). Therefore, the presence of InGaN is an essential condition for obtaining a high level of light emission efficiency, and as the reason therefor, localization of the energy state caused by fluctuations of the content of In in the InGaN quantum well layer is considered as being an essential activity. Furthermore, Mukai et al. (for example, refer to Non-Patent Document 6) manufactured a prototype of an LED of UV, blue, and green with the same structure, and investigated the electric current at a peak wavelength and temperature dependency. As a result, they have confirmed that if the peak wavelength is 375 nm or greater, it shifts to the shorter wavelength side when raising the electric current, and there is no variation in regards to temperature. In contrast, a GaInAlP based LED has an opposite effect in which its peak wavelength does not change even if the electric current is raised, but it shifts to the longer wavelength side if the temperature is raised, and therefore the light-emitting mechanism of these are considered completely different (for example, refer to Non-Patent Document 7).
Moreover, in 1996, Chichibu et al. manufactured a prototype of an LED using quantum wells in which x in undoped InxGa1-xN is changed to x=0.2, 0.3, and 0.45, and they conducted energy structure analysis from an optical point of view in various types of methods in which: electric current was applied thereto to cause it to emit light, and the composition dependence thereof at the light emission peak was measured; and the composition dependence thereof was measured when homogeneous light with an altered wavelength was irradiated thereon to measure the electromotive force. Moreover, they discovered in the sectional TEM of the MQW that portions with darker colors were cyclically present in the well layer. Based on the above results, they summarized that a minimum potential was present due to the presence of fluctuation in the In concentration, and consequently holes locally existed in the minimum potential portion. As a result, light emission efficiency was increased (for example, refer to Non-Patent Document 8). After this, research has progressed systematically, and it is considered that the excitation life time becomes longer due to the local existence of an exciter in the quantum well layer, the diffusion distance of holes becomes shorter than the intervals of threading dislocation, and therefore the presence of threading dislocations, which are the non-radiative centers, does not have an influence on brightness (for example, refer to Non-Patent Document 9 to 14). Moreover, in 1998, Nakamura summarized the above research, organizing roles of crystal defects in GaN based LED and LD, and he concluded that improvement in crystallinity has no relation to light emission efficiency in an LED (for example, refer to Non-Patent Document 15).
Moreover, in 1997, Usui et al. proposed an epitaxially laterally overgrown method (for example, refer to Non-Patent Document 16), and further improvement in crystallinity became possible. In this method, having grown GaN by means of a MOCVD method with use of a low temperature buffer method, the surface is partially covered with SiO2 by means of a photolithography technique. If GaN is grown by means of a HPVE method after this, GaN can no longer grow on the SiO2, and therefore it grows laterally and is filled on the SiO2. This type of GaN which has grown in a lateral direction has an extremely low level of threading dislocation, and therefore GaN single crystals with an extremely low level of threading dislocation can be formed only on the SiO2 on a wafer.
Furthermore, in 1998, Mukai, Takegawa, and Nakamura reported results of manufacturing an LED in a structure of Mg doped p-GaN 2000 Å/Mg doped p-AlGaN 300 Å/undoped InGaN 25 Å/undoped GaN 2000 Å/Si doped n-GaN 2.3 μm/undoped GaN 1.5 μm/ as an ELOG substrate, and comparing it with an LED in which LT GaN buffer 300 Å is laminated on sapphire and the above structure is laminated thereon (for example, refer to Non-Patent Document 17). Here, the threading dislocation density could be reduced from 1×10+10/cm2 to 7×10+6/cm2. However, the light emission outputs are respectively 20 mA at 6 mW and were substantially equal while the peak wavelengths were slightly different at 470 nm and 464 nm. However, while the leakage current at low voltage 1.8V was 0.001 μA or lower with an ELOG product, which has a high level of dislocation density, it was 0.01 μA or greater with an m-sapphire product.
Moreover, in 1999, Chichibu et al. reported that they grew an SQW of InGaN on a substrate which used ELOG, and measured time integration and time-resolved PL at a portion with high dislocation density and at a portion with low dislocation density (for example, refer to Non-Patent Document 18). As a result of this, it was confirmed that optical characteristics do not change even if changes occur in threading dislocation density. Here, there was a discussion which questions whether penetrating dislocation is not the non light-emitting center. However, Sugahara et al. confirmed and reported that a dislocation is the non light-emitting center even in a GaN based LED, by observing the same location using TEM and CL (for example, refer to Non-Patent Document 30).
Moreover, in 1999, Mukai and Nakamura manufactured an LED structure of UV, blue, and green on both of a GaN which uses ELOG and on a general sapphire, and compared light emission efficiency (for example, refer to Non-Patent Document 19). As a result, in the case of using GaN as an active layer, the level of light emission efficiency was higher when the level of crystallinity was higher with use of ELOG. However, in the case of using InGaN as an active layer, the level of light emission efficiency did not change even with ELOG or with an LED grown on sapphire. However, in the case of UV, the level of light emission efficiency was higher with ELOG when the electric current density was raised. As the reason for this result, local existence of an exciter caused by fluctuations in the In concentration described above was given.
Furthermore, in 2006, there was a report of summarizing the above discussion which described that this is because InN locally captures a hole and the distance of the hole moving until the moment of light emission is extremely short, and therefore, light is emitted before the hole falls into the defect (for example, refer to Non-Patent Document 20). Therefore, as a direction of development in increasing the level of brightness of an LED, it was proposed that artificially creating inhomogeneity in a quantum well was preferred. This report was made by both Nakamura and Akasaki, who had led the development and academic research in the area of GaN based LED and LD, and it may be understood that this is the terminal point of current achievement in the area.
Moreover, there was invented a method for manufacturing GaN single crystal crystals such that when forming a GaN single crystal film on a sapphire substrate, approximately 300 Å of AlN or GaN is film-formed at a low temperature between 500 and 800° C., an island-shaped crystal is formed by raising the temperature thereafter, and the lateral direction growth of this island-shaped crystal is used. And this method became widely prevailed.
This layer, which was grown at a temperature lower than the growing temperature, is called a buffer layer. This layer is polycrystalline or amorphous, and only a part thereof becomes crystallized at a high temperature growth and does not entirely become crystallized. By partially crystallizing a considerable number of locations in this way, there is exerted a mechanism in which only crystals with a matched crystal orientation unite with each other in the lateral direction, thereby bonding crystals of sapphire and GaN. GaN single crystal growth has become possible with this type of low temperature buffer layer technology, and the market of high-brightness blue LEDs has rapidly expanded.
Although GaN single crystal growth has become possible with the low temperature buffering technique, the level of defect density in the high-brightness blue LED is extremely high as a crystal, compared to other single crystals such as GaAs. A widely accepted theory is that defect density and light emission efficiency has no correlation therebetween as long as there is a certain level of crystallinity, since an LED of high level output can be manufactured even if the level of defect density in GaN crystals is comparatively high, and a considerable number of published documents have discussed this (for example, refer to Non-Patent Documents 15 and 20). The assertion of these respective documents can be summarized as follows.
It became possible to control the threading dislocation density by lateral epitaxial coating growth, and create portions each having different levels of dislocation density. Consequently, it became possible to evaluate its relationship with brightness.
As a result, having investigated the relationship between photoluminescence (PL) and penetrating dislocation density in an InGaN quantum well, it has been revealed that light emission intensity is not dependent on penetrating dislocation density. Although penetrating dislocation serves as a non light-emitting recombination channel of surrounding carriers present within a diffusion length, it only reduces the volume of a light-emitting region. This is because, a minimal potential is generated in an InGaN multiple quantum well due to the fluctuations in the In concentration, and a hole is captured therein. As a result, light is emitted before the hole moves to the dislocation, which is a non light-emitting center.
Therefore, the development of improving light emission efficiency of a GaN based LED is being oriented towards improving light extraction efficiency, and research and development for increasing light emission efficiency by reducing defect density has been hardly conducted for a long time. Lateral epitaxial coating growth has been employed mainly in LDs. However, in 2001, Okagawa et al. had success in lateral growth by conducting a concave-convex treatment on a sapphire substrate, and they applied this to an LED (for example, refer to Non-Patent Document 21). Here, by reducing the dislocation density from 4×10+8/cm2 to 1.5×10+8/cm2, output improved from 3.5 mW to 9 mW with UV at a peak wavelength 382 nm. However, later, the output improvement of Non-Patent Document 21 was interpreted as being the increase in the light extraction efficiency due to the unevenness treatment if the sapphire surface. Accordingly, it is thought that such a technical recognition that the level of brightness can be increased by improving the level of crystallinity, is completely disappeared in the area of LEDs in recent years.
However, it is clear that if light emission efficiency can be further improved, applications in the area of LEDs will dramatically expand, and consequently, there has been an increasing demand for improvement in light emission efficiency. Light emission efficiency of an LED can be generally separated into two concepts, namely internal quantum efficiency and external quantum efficiency. Internal quantum efficiency is an indicator of how much of applied electric current has become light, and external quantum efficiency is an indicator of how effectively generated light can be used. Although it can be conceptually separated in this way, in reality, there is a loss such that the light which is once generated is absorbed into an energy gap of GaInN. Further, there is a problem as for this type of light loss in which it is fundamentally difficult to determine whether it has once become light or not. If the theory, in which light emission efficiency will not change even if crystallinity is improved, is accepted, it means that internal quantum efficiency has reached the limit. Therefore, in reality, the majority of recent research and development for improving light emission efficiency is aiming to increase light extraction efficiency to thereby improve external quantum efficiency.
[Prior Art Documents]
[Patent Documents]
[Patent Document 1] Japanese Patent No. 3026087
[Patent Document 2] Japanese Unexamined Patent Application, First Publication No. H04-297023
[Non-Patent Documents]
[Non-Patent Document 1] P. M. Petroff, O. G. Lorimor, J. M. Raiston, “Defect structure induced during forward-bias degradation of GaP green-light-emitting diodes”, J. Appl. Phys. Vol. 47 (1976) 1583
[Non-Patent Document 2] S. Guha, J. M. DePuydt, M. A. Haase, J. Qiu, H. Cheng “Degradation of II-VI based blue-green light emitters”, Appl. Phys. Lett. 63 (1993) 3107
[Non-Patent Document 3] T. Sugahara, H. Sat, M. Hao, Y. Naoi, S. Kurai, S. Tottori, K. Yamashita, K. Nishino, L. T. Romano, S. Sakai “Direct Evidence that Dislocations are Non-Radiative Recombination Centers in GaN”, Jpn. J. Appl. Phys. Vol. 37 (1998) L398
[Non-Patent Document 4] S. J. Rosner, E. c. Carr, M. J. Ludowixe, G. Girolami, H. I. Erikson “Correlation of cathodoluminescence inhomogeneity with microstructural defects in epitaxial GaN grown by metalorganic Chemical-vapor deposition”, Appl. Phys. Lett. 70 (1997) 420
[Non-Patent Document 5] T. Mukai, D. Morita, S. Nakamura, “High-Power UV InGaN/AlGaN double-heterostructure LEDs”, J. Crystal Growth Vol. 189/190 (1998) p778
[Non-Patent Document 6] S. Nakamura, M. Senoh, N. Iwasa, S. Nagahawa, “Superbright Green InGaN Single-Quantum-Well-Structure Light Emitting Diodes”, Jpn. J. Appl. Phys. Vol. 34 (1995) pL1332
[Non-Patent Document 7] T. Mukai, M. Yamada, S. Nakamura, “Current and Temperature Dependence of Electroluminnescence of InGaN-Based UV/Blue/Green Light-Emitting Diodes”, Jpn. J. Appl. Phys. Vol. 37 (1998) L1358
[Non-Patent Document 8] S. Chichibu, T. Azuhata, T. Sota, S. Nakamura, “Spontaneous emission of localized excitons in InGaN single and multiquantum well structures”, Appl. Phys. Lett. Vol. 69 (1996) 4188
[Non-Patent Document 9] S. Chichibu, A. Shikanai, T. Azuhata, T. Sota, A. Kuramata, K. Horino, S. Nakamura, “Effects of biaxial strain on exciton resonance energies of hexagonal GaN heteroepitaxial layers”, Appl. Phys. Lett. Vol. 68 (1996) 3766
[Non-Patent Document 10] S. Chichibu, T. Azuhata, T. Sota, S. Nakamura, “Excitonic emissions from hexagonal GaN epitaxial layers”, J. Appl. Phys. Vol. 79 (1996) p2784
[Non-Patent Document 11] S. Chichibu, Wada, S. Nakamura, “Spatially resolved cathodoluminescence spectra of InGaN quantum wells”, Appl. Phys. Lett. Vol. 71 (1997) p2346
[Non-Patent Document 12] Y. Narukawa, Y. Kawakami, S. Fujita, S. Fujita, S. Nakamura, “Recombination dynamics of localized excitons in In0.20Ga0.80N—In0.05Ga0.95N multiple quantum wells”, Phys. Rev. B Vol. 55 (1997) R1938
[Non-Patent Document 13] Y. Narukawa, Y. Kawakami, M. Funato, S. Fujita, S. Fujita, S. Nakamura, “Role of self-formed InGaN quantum dots for exciton localization in the purple laser diode emitting at 420 nm”, Appl. Phys. Lett. Vol. 70 (1997) 981
[Non-Patent Document 14] S. Chichibu, T. Sota, K. Wada, S. Nakamura, “Exciton localization on InGaN quantum well devices”, J. Vac. Sci. Technol. B Vol. 16 (1998) p2204
[Non-Patent Document 15] S. Nakamaura, “The Roles of Structural Imperfections in InGaN-Based BlueLight-Emitting Diodes and Laser Diodes”, Science 14 Aug. 1998 vol. 281 No. 5379 pp 956-951
[Non-Patent Document 16] A. Usui, H. Sunakawa, A. Sakai, A. Yamaguchi, “Thick GaN Epitaxial Growth with Low Dislocation Density by Hydride Vapor Phase Epitaxy”, Jpn. J. Appl. Phys. Vol. 36 (1997) L899
[Non-Patent Document 17] T. Mukai, K. Takekawa, S. Nakamura, “InGaN-Based Blue Light-Emitting Diodes Grown on Epitaxially Laterally Overgrown GaN Substrates”, Jpn. J. Appl. Phys. Vol. 37 (1998) L839
[Non-Patent Document 18] S. F. Chichibu, H. Marchand, M. S. Minsky, S. Keller, P. T. Fini, J. P. Ibbetson, S. B. Fleischer, J. S. Speck, J. E. Bowers, E. Hu, U. K. Mishra, S. P. DenBaars, T. Deguchi, T. Sota, S. Nakamura, Appl. Phys. Lett. Vol. 74 (1999) 1460 “Emission mechanisms of bulk GaN and InGaN quantum wells prepared by lateral epitaxial overgrowth”
[Non-Patent Document 19] T. Mukai, S. Nakamura, Jpn. J. Appl. Phys. Vol. 38 (1999) 5735 “Ultraviolet InGaN and GaN Single-Quantum-Well-Structure Light-Emitting Diodes Grown on Epitaxially Laterally Overgrown GaN Substrates”
[Non-Patent Document 20] CHICHIBU Shigefusa F., UEDONO Akira, ONUMA Takeyoshi, KOYAMA Takahiro (Univ. Tsukuba, Tsukuba, JPN), HASKELL Benjamin A., FINI Paul T., DENBAARS Steven P., SPECK James S., NAKAMURA Shuji (JST-ERATO, Kawaguchi, JPN), UEDONO Akira (National Inst. Materials Sci., Tsukuba, JPN), CHAKRABORTY Arpan, KELLER Stacia, MISHRA Umesh K., (Univ. California, California, USA), YAMAGUCHI Shigeo, KAMIYAMA Satoshi, AMANO Hiroshi, AKASAKI Isamu (Meijo Univ., Nagoya, JPN), HAN Jung (Yale Univ., Connecticut, USA), SOTA Takayuki (Waseda Univ., Shinjuku, JPN) “Origin of defect-insensitive emission probability in In-containing (Al, In, Ga) N alloy semiconductors”, Nature Materials Vol. 5, No. 10, Page 810-816 (2006.10)
[Non-Patent Document 21] OKAGAWA Hiroaki, OUCHI Yoichiro, TSUNEKAWA Takashi, TADATOMO Kazuyuki, KATO Munehiro “Developing high level output UV LED, using LEPS method”, Mitsubishi Densen Kogyo Jiho (Mitsubishi Cable Industries, Inc. Times) No. 98 (2001), 92.